Markov Switching Model

[McFeeney]

Hi Guys,

I have recently started using eview for a project in college. I am using Hamilton's 1989 model that was modified in 2002 by Garcia and Schaller to try and show that interest rates have a different effects dependent on what state the interest rate change happens in. So I have got the Gnp data for the US and preformed a 100*dlog(gnp) on it and let the series equal g. I have got the Federal funds rate and and preformed a dlog(fed) on it and let the series equal f. Please correct me if I have made any mistakes so far. I then go to quick/estimate equation. I change the method to switchreg. I put g followed by c in the dependent variable box. ar(1) ar(2) ar(3) ar(4) goes in the non-switching box. This is the part I'm not sure about. Where do I put the interest rate? I have put it in the Probability regressioners box after c like f(-1) f(-2) f(-3) f(-4). Is that correct? When I get the filters graph it looks correct. I'm also unsure how to interpret the data. What coefficient means a change in the interest rate leads to a change in the GNP in either state?

Any help would be appreciated. Below is the output I got.
Dependent Variable: G
Method: Markov Switching Regression (BFGS / Marquardt steps)
Date: 02/11/17 Time: 13:57
Sample (adjusted): 1957Q2 2015Q1
Included observations: 232 after adjustments
Number of states: 2
Initial probabilities obtained from ergodic solution
Standard errors & covariance computed using observed Hessian
Random search: 25 starting values with 10 iterations using 1 standard
deviation (rng=kn, seed=73038125)
Convergence achieved after 24 iterations


Variable Coefficient Std. Error z-Statistic Prob.


Regime 1


C 1.704810 0.201872 8.444991 0.0000


Regime 2


C -0.260794 0.410985 -0.634557 0.5257


Common


AR(1) 0.365804 0.108313 3.377300 0.0007
AR(2) 0.250399 0.087213 2.871108 0.0041
AR(3) -0.029693 0.101192 -0.293430 0.7692
AR(4) 0.166981 0.082094 2.034020 0.0419
LOG(SIGMA) -0.343601 0.058422 -5.881417 0.0000


Transition Matrix Parameters


P11-C 3.496836 0.738764 4.733361 0.0000
P11-F(-1) 5.107281 3.419715 1.493481 0.1353
P11-F(-2) -6.216290 2.640993 -2.353770 0.0186
P11-F(-3) 0.642739 1.314317 0.489029 0.6248
P11-F(-4) -3.265967 2.051628 -1.591891 0.1114
P21-C -1.409842 1.738490 -0.810958 0.4174
P21-F(-1) -0.252743 3.381748 -0.074737 0.9404
P21-F(-2) -11.79836 10.05561 -1.173312 0.2407
P21-F(-3) 2.206024 6.362819 0.346705 0.7288
P21-F(-4) 8.843965 10.22345 0.865066 0.3870


Mean dependent var 1.567927 S.D. dependent var 0.993521
S.E. of regression 0.840663 Sum squared resid 161.8374
Durbin-Watson stat 2.149667 Log likelihood -280.2760
Akaike info criterion 2.519288 Schwarz criterion 2.768801
Hannan-Quinn criter. 2.619869


Inverted AR Roots .87 .08+.53i .08-.53i -.66






Equation: UNTITLED
Date: 02/11/17 Time: 14:10
Transition summary: Time-varying Markov transition
probabilities and expected durations
Sample (adjusted): 1957Q1 2015Q1
Included observations: 233 after adjustments


Time-varying transition probabilities:
P(i, k) = P(s(t) = k | s(t-1) = i)
(row = i / column = j)
1 2
Mean 1 0.896996 0.103004
2 0.044264 0.955736

1 2
Std. Dev. 1 0.304619 0.304619
2 0.155417 0.155417



Time-varying expected durations:

1 2
Mean NA 2.34E+10
Std. Dev. NA 3.57E+11